When monolithic applications are decomposed into microservices, the architecture must rely on network protocols to communicate between the distributed components. Therefore, when scaling microservices it is necessary to scale the entire application stack of networking protocols and microservices logic.
The Reaktivity Project uses shared memory streams to directly communicate between each different layer in the application stack, including both the networking protocols and the microservices logic. Parallel execution of each layer in the application stack takes full advantage of the available CPU cores to maximize throughput with predictably low latency.
Recognizing that large systems can have 1000s of CPU cores, this technology can be deployed economically in the cloud to provide affordable real-time solutions.